Long noncoding RNAs (lncs) as sinks in Prader Willi Syndrome

Abstract

In this issue of Molecular Cell, Yin et al. (2012) identify a class of long noncoding RNAs (lncRNAs) and propose a new mechanism as to how they contribute to the pathogenesis of Prader-Willi Syndrome.

Prader-Willi syndrome (PWS) is a neurogenetic disorder that has been mapped to an imprinted region of chromosome 15. Most PWS patients have a de novo deletion of a 5–6 Mb region on the paternally derived chromosome 15, resulting in a complete loss of expression of the paternally expressed genes from this locus (Cassidy et al., 2012). There are several coding genes, including one expressed exclusively from the paternal chromosome in the brain, a cluster of box C/D small nucleolar RNAs (snoRNAs) and several antisense transcripts expressed from this region. While the role of each expressed gene in the pathogenesis of PWS remains relatively unknown, there is strong evidence that the SNORD116 snoRNA gene cluster plays a major role (Sahoo et al., 2008). Box C/D snoRNAs usually function as guides for ribose methylation of an RNA target, yet no target has been confirmed for any of the SNORD116 snoRNAs. Their role in PWS has remained elusive until now. In this issue of Molecular Cell, Yin et al. (2012) describe a class of lncRNAs derived from the SNORD116 snoRNA cluster that are implicated in PWS pathogenesis.

Deep sequencing of non-polyadenylated RNAs in HeLa cells and human embryonic stem (ES) cells uncovered a number of long noncoding RNAs (lncRNAs) that are derived from introns (Yang et al., 2011). Further analysis of these RNAs revealed a class of lncRNAs, sno-lncRNAs, that contain a snoRNA at each of their 5′ and 3′ ends. sno-lncRNAs range in size from 1,200–2,900 nts. They are processed using the snoRNP biogenesis machinery that requires the conserved box C/D snoRNA motifs. Intriguingly, in human ES cells, at least five sno-lncRNAs are abundantly expressed from the imprinted region on chromosome 15 implicated in PWS. These snolncRNAs are retained in the nucleus, co-localize, and accumulate near their site of transcription.

Crosslinking and immunoprecipitation followed by deep sequencing from human ES cells showed that SNORD116 transcripts are strongly bound by Fox2, an alternative splicing regulator. Fox2 recognizes and binds five different hexanucleotide motifs (Yeo et al., 2009). Each sno-lncRNA contains multiple Fox2 binding sites, suggesting that the function of sno-lncRNAs lies in regulating pre-mRNA splicing. Indeed, when Yin and colleagues analyzed splicing patterns of Fox2 regulated cassette exons in the presence and absence of sno-lncRNAs they saw consistent changes in splicing. Additionally, the absence of Fox2 affected the splicing patterns in the opposite way. Taken together, these results suggest a potential pathogenic mechanism for PWS (Figure 1). In normal pluripotent cells, sno-lncRNAs are expressed from the paternal chromosome where they bind Fox2, effectively titrating its levels and maintaining splicing patterns that promote normal development and growth. In patients with PWS, the locus is deleted, and thus the sno-lncRNAs are not expressed. PWS may arise because higher effective Fox2 levels lead to altered pre-mRNA splicing patterns and dysregulation of development. sno-lncRNAs transcribed from the imprinted Prader-Willi locus of chromosome 15 thus act as an RNA sink for the alternative splicing factor, Fox2.

Figure 1.

Long non-coding RNAs (lncs) as sinks in Prader Willi Syndrome.

sno-lncRNAs are normally expressed from the SNORD116 locus in cells containing an intact paternal chromosome 15. These RNAs localize in small subnuclear speckles where they sequester the splicing regulator Fox2. This effectively reduces the levels of free Fox2 and maintains pre-mRNA splicing patterns required for normal development. Conversely, in patients with Prader-Willi Syndrome, the SNORD116 locus is deleted and the sno-lncRNAs are not expressed. In the absence of sno-lncRNAs, free Fox2 levels increase, causing changes in pre-mRNA splicing that may lead to abnormal development and disease. Thus, the sno-lncRNAs from the SNORD116 locus may contribute to PWS pathogenesis by serving as molecular sinks for Fox2. This cartoon representation accurately depicts one possible molecular scenario based on the model proposed by Yin et al. (2012).

Identification and cataloguing of the lncRNAs of the human genome has revealed a number of commonalities between lncRNAs and sno-lncRNAs. The vast majority of both lncRNAs and sno-lncRNAs have minimal protein coding potential. Compared to protein coding mRNAs, lncRNAs are significantly enriched in the nucleus and in the poly(A)⁻ fraction in all cell types (Derrien et al., 2012). Similarly, sno-lncRNAs are poly(A)⁻ nuclear transcripts (Yin et al., 2012). Despite their novelty, sno-lncRNAs share many characteristics with known lncRNAs.

There are a few key differences, though, that set sno-lncRNAs apart. The sno-lncRNAs from the Prader-Willi locus are expressed at levels similar to some abundant histone mRNAs and are 1,200–2,900 nts in length (Yin et al., 2012). In contrast, many lncRNAs are expressed at much lower levels and are generally shorter than protein coding genes (Derrien et al., 2012). Interestingly, only the snoRNA sequences of the sno-lncRNAs are conserved among placental mammals (UCSC Genome Browser), while most lncRNAs are evolutionarily conserved across the majority of placental mammals. The results obtained by Yin et al. (2012) suggest that the Prader-Willi sno-lncRNAs retain their snoRNA “caps” or ends, which, in turn, promote the stability of the sno-lncRNAs. Conversely, many lncRNAs serve as precursors for small RNAs (Askarian-Amiri et al., 2011). These differences between sno-lncRNAs and lncRNAs may provide insights into the molecular significance of this new lncRNA class.

sno-lncRNAs also differ in their proposed function from known lncRNAs, which have diverse roles in modulating gene expression (Wilusz et al., 2012). A number of other lncRNAs, including Air and Xist, are also expressed from imprinted loci. However, these lncRNAs play a role in establishing or maintaining a specific epigenetic state (Amaral et al., 2011), with no known role in alternative pre-mRNA splicing.

While the study by Yin et al. (2012) defines a novel class of lncRNAs and adds some clarity to the mystery of PWS pathogenesis, it also raises a number of interesting questions. Yin et al. (2012) showed that sno-lncRNAs co-localize with Fox2 and form subnuclear speckles near their chromosomal locus. They propose that splicing patterns are altered specifically in these subnuclear neighborhoods. Yet, the significance of this localization is not clear. Is this localization important solely for sno-lncRNA biogenesis? Or is it critical for optimal sno-lncRNA function? Furthermore, it is not known if the genes that reside in these neighborhoods are transcribed into pre-mRNAs that are alternatively spliced in a Fox2-dependent manner. It will also be important to determine if Fox2 is the only protein that binds sno-lncRNAs as it is possible that sno-lncRNAs interact with multiple proteins to carry out distinct functions within the nucleus. Further analysis of the composition of the subnuclear speckles may provide clues to other potential binding partners and functions of sno-lncRNAs. Finally, Yin et al. (2012) have focused only on the sno-lncRNAs expressed from the Prader-Willi region on chromosome 15. Thus, it will be of great importance to identify and characterize other sno-lncRNAs to determine if the properties of the Prader-Willi sno-lncRNAs are specific or universal traits of this novel class of lncRNAs.